NVIDIA publishes GR00T N2 ONNX export for SSD pick-and-place on Hugging Face
Overview
NVIDIA publishes a native split ONNX export of GR00T N2 checkpoint 18200 on Hugging Face for SSD pickup and placement tasks.
The checkpoint was trained on 196 pickup and 149 placement episodes, with 310 training and 35 validation episodes, and one shared set of graphs serves both tasks through a host-side task selector.
NVIDIA says the package is not a TensorRT engine or a robot-ready policy.yaml, so users must build and validate it on the target platform before testing. NVIDIA does not assert full ONNX/eager numerical parity, TensorRT validation, or robot success rates.
Written by AI from the articles below · updated Oct 9, 9:03 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- NVIDIA · new models on Hugging FaceOfficialNVIDIA releases GR00T N2 ONNX checkpoint for SSD pick-and-place tasks
AINVIDIA publishes a GR00T N2 checkpoint, step 18200, as a native split ONNX export on Hugging Face for SSD pickup and placement. The model trained on 196 pickup and 149 placement episodes, with 310 training and 35 validation episodes, and one shared set of graphs serves both tasks through a host-side task selector. The package is not a TensorRT engine or a robot-ready policy, and NVIDIA does not assert full ONNX/eager numerical parity or robot success rates.
Heat trend
Not enough continuous observations to show a trend yet.